Trajectory Bundle Clustering for Map Data Analysis
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Mapping-related service providers face challenges in organizing and analyzing high-volume probe data to create and update map data, especially at complex locations like intersections and highway interchanges, due to inaccuracies in geo-positioning and difficulties in defining polygons for trajectory grouping.
Innovation Solution
A system that automatically groups trajectories within a bounded geographic area by computing similarities of curves represented by probe data, eliminating the need for map matching and reducing errors, using methods like discrete Frechet distance and dynamic time warping to cluster trajectories into bundles representing possible maneuvers.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If probe data is organized using traditional polygon-based grouping methods, then map data can be generated, but inaccuracies in geo-positioning and errors in defining polygons reduce the precision of trajectory analysis
Solution Approach 1:
The patent introduces trajectory bundles as an intermediary structure between raw probe data and map data. Instead of directly using polygon-based grouping which is sensitive to geo-positioning errors, the system creates trajectory bundles that represent probable maneuver paths. This intermediary layer absorbs the inaccuracies in individual probe points while maintaining the overall structural integrity of the maneuver representation, thereby resolving the contradiction between measurement precision and reliability.
Solution Approach 2:
The system creates simplified copies of complex trajectory data by representing multiple similar probe trajectories as a single trajectory bundle. Rather than analyzing each individual probe point which suffers from positioning errors, the system creates a consolidated representation that captures the essential maneuver pattern. This copying approach filters out noise from individual measurements while preserving the fundamental trajectory characteristics.
2Quantity of substance
If probe data volume is increased to improve map data coverage, then more complete map analysis is achieved, but the complexity of organizing and processing the data increases significantly
Solution Approach 1:
The patent merges multiple individual probe trajectories into consolidated trajectory bundles based on similarity criteria. By combining numerous individual data streams that would otherwise require separate processing, the system reduces computational complexity while maintaining comprehensive coverage. The merging process groups trajectories that represent the same maneuver type, allowing the system to process large volumes of probe data efficiently through batch operations on bundled trajectories rather than individual trajectory processing.
Solution Approach 2:
The system segments the large volume of probe data into manageable trajectory bundles organized by geographic area and maneuver type. This segmentation divides the overwhelming task of processing all probe data into smaller, independent units that can be processed in parallel. Each trajectory bundle represents a discrete maneuver at a specific location, allowing the processing system to handle large data volumes through systematic division and parallel processing of segmented data units.
3Ease of manufacture
If map matching is used to organize trajectories, then trajectories can be aligned with existing road networks, but the process introduces additional errors and dependencies on underlying map topology
Solution Approach 1:
The patent inverts the traditional map matching approach by not forcing probe trajectories to conform to existing map road networks. Instead of adjusting trajectories to match predefined roads, the system allows trajectory bundles to emerge naturally from the probe data itself, creating a data-driven representation of maneuvers. This inversion eliminates the circular dependency where trajectories must match roads that were originally created from similar trajectories, thereby improving accuracy by removing the bias introduced by pre-existing map topology.
Data Source
Figure 1
Figure 2
Figure 3
AI summary
An approach is provided for generating trajectory bundles for map data analysis. The approach involves receiving probe data associated with the bounded geographic area. The probe data are collected from sensors of a plurality of devices traveling in the bounded geographic area, and includes probe points indicating a position, a heading, a speed, a time, or a combination thereof. The approach also involves constructing a plurality of trajectories from the probe points to represent respective movement paths of said each of the plurality of devices. The approach further involves computing similarities among a plurality of curves represented by the plurality of trajectories. The approach further involves clustering the plurality of trajectories into trajectory bundles based on the similarities with each bundle representing a possible maneuver within the bounded geographic area. The approach further involves generating a map of the bounded geographic area based on the trajectory bundles.